Abstract: In order to solve the problem of uncertainty processing and the fuzziness of index weight in the risk analysis of gas explosion, a risk assessment model of gas explosion using the Bayesian network and game theory is proposed. Firstly, the topological structure model of gas explosion is constructed based on expertise and accident-causing theories. At the same time, the subjective and objective weights of experts are determined by game theory optimization. Then, the prior probability and conditional probability of risk factors are calculated based on the fuzzy set theory. Secondly, the Bayesian inference technology is used to calculate the probability of gas explosion and the posterior probability distribution of risk factors. Finally, the sensitivity analysis and key cause chain analysis are used to find out the key risk factors and key risk paths affecting gas explosion. The case study shows that the probability of gas explosion in the working face of a coal mine in Shandong Province is 6.5%. The main ignition sources of gas explosion are electric failures, friction between rocks, blasting sparks and coal spontaneous combustion, while the main causes of gas accumulation are ventilation system failures and ventilation system design. Through sensitivity analysis, it can be seen that two factors of ventilation system failures and electric failures are the most sensitive. According to the analysis of the key causal chain, it shows that the combination of the path of electric failures→ventilation system failures→ventilation measures→gas control measures→gas transfinite and the path of electric failures→electric spark→ignition source leads to the highest possibility of gas explosion. Compared with normal conditions, the probability of gas explosion increased by 301.5%. At the same time, the combination weighting method based on game theory has comprehensive advantages in improving the objectivity and accuracy of index weights, and the evaluation results accords with the actual situation of coal mines.